The Charge Transfer to Solvent Spectrum of Iodide ... - ACS Publications

Introduction. CTTS spectra, exhibited in the UV range by a number of anions ..... a critical parameter, and changes of the cavity radius with temperat...
1 downloads 0 Views 1018KB Size
114

A. Barkatt and C. A. Angell

The Charge Transfer to Solvent Spectrum of Iodide in Supercooled Water and Glass-Forming Aqueous Solutions A. Barkatt and C. A. Angell” Chemistry Department, Purdue UniversiW, West Lafayetto, Indiana 47907 (Received April 16, 1976) Publication costs assisted by the National Science Foundation

Measurements of the peak position and the line shape of the charge transfer to solvent (CTTS) band of aqueous I- were carried out using thin samples squeezed between quartz disks. Results obtained in glass-forming aqueous magnesium acetate (Mg(0Ac)J solutions show a temperature dependence similar to that observed in pure water above Tg, but below Tgboth peak position and bandwidth suddenly become much less temperature dependent. This is a direct indication of the importance of configurational,as opposed to vibrational, degrees of freedom in accounting for temperature effects on CTTS transitions. In pure water supercooling to -31 “C was achieved. Below 0 “C, the temperature dependence of the energy of the transition deviates from linearity and becomes gradually smaller, and changes in bandwidth are arrested. These effects are discussed in terms of water structure in the supercooled region. Correlations of structural effects on CTTS with data concerning the absorption of the hydrated electron are discussed. Results are also given concerning the effects of solutes (e.g., I-, OAc-, Br-) on the CTTS band. In particular, systematic observations were made on the line shape as a function of solute concentrations. The results are discussed using the physical picture of the diffuse model, attributing most structural effects to changes in the ground state energy.

Introduction CTTS spectra, exhibited in the UV range by a number of anions, are associated with a transition from a ground state in which the negative charge resides mainly on the central ion surrounded by a solvation shell to an excited state in which negative charge is transferred to the solvating medium. The excited state can be regarded, in first approximation, as a combination of a neutral radical and a solvated electron. Since the ground state is a solvated ion its energy is considerably influenced by the solvent, although it is defined principally by the central ion. On the other hand, in defining the electronic excited state the solvent plays an integral part. Most spectroscopic transitions observed in solution are influenced to some extent by the solvent, especially in cases such as the symmetry-sensitive ligand field d-d transitions where the environment determines the symmetry of the field sensed by the ion. However, CTTS transitions constitute an unusually environment-sensitive class of transitions, and consequently are useful as probes of changes in solvent structure with temperature, pressure, or the addition of another solute or a cosolvent. The dependence of CTTS spectra on environmental parameters shows many similarities to the behavior of transitions between energy levels of the solvated electron itself.’p2 Indeed, the excited state of the CTTS transition closely approximates the structure of the solvated electron in its ground state, and data obtained for the energy levels of the solvated electron have played a major role in the development of models for CTTS transitions. A comprehensive review of experimental findings and theoretical models for CTTS transitions was published by Blandamer and Fox.’ Most studies have been concerned with solvent, second solute, and/or temperature effects on the position of the band maximum for the lower energy (E,,, x 226 nm) of the two CTTS bands, with the aim of providing the empirical basis for a successful theoretical modeling of the transition. Unfortunately, despite the large amount of data on conventional solutions now available, the models so far proposed have been only The Journal of Physical Chemistry, Vol. 81, No. 2, 1977

moderately successful. Using a “polaron” model,3 with rather crude approximationsfor the excited state, the value of E, can be understood, but the temperature dependence is not properly accounted for. More recent models, “diffuse”,4 “~onfined”,~ and “configuration coordinate”> to be discussed later, have had interpretation of the temperature dependence as one of their primary concerns, but have also had only limited success. The aims of the present investigation are twofold and self-reinforcing. In the first place we seek to sharpen the understanding of the nature and origin of the temperature dependence of the CTTS bands by extending their study to glass-forming solutions in which we can experimentally separate, by measurements above and below the glass transition temperature, the effects on dE,,/dT of purely vibrational from those of structural (configurational) e~pansion.~ With the resultant clarification of the role of solution structure on the transition energy, we are then able to employ the structure sensitivity of the Cl”S band transition to obtain some insight into structural ordering processes in water in the low-temperature supercooled regime in which recent thermodynamic and transport studies have revealed highly anomalous behavior.*pg

Experimental Section The study of supercooled water requires small sample sizes, suggesting the use of thin aqueous layers enclosed between quartz surfaces for spectral studies. This is a method which was first developed by Fromherz and Menschick’O in their pioneering work on CTTS spectra in high concentrations of aqueous halides. More recently it was considered preferable to carry out such studies in 0.1-1-cm cells and to observe only the low-energy tail of the band,’l a method which cannot distinguish between shifts of the maximum and effects on the bandwidth. The use of thin samples makes it possible to extend the studies of solute effects on the entire band to the numerous cases of solutes with a significant absorption at high concentrations in the 200-250-nm range, such as the acetate ion, OAc- (which was used in the present work to obtain

CTTS Spectrum of Iodide

TABLE I: Values of E,, Solution

115

(20 "C) and Average dE,,/dT Tg, "C

Water Water Water Water Water

R = 14(3.105 M ) Mg(OAc),

-81

R=7(4.957M) Mg(OAc), 2 M Li,SO, 2 M Li,SO, 2 M MgSO, 2 M MgSO, 4 M NH,C1 4 M NH4Cl 4 M NH,Br

-42

5x 4.28 X 10'' 1 x 10-4 2 x 10-5 0.1

for Iodide in Water and Aqueous Solutions Av dE,,/dT, kcal mol-' Temp range, "C kcal mol-' deg-' 0.0328 126.40 1-91 18-63 0.03320 126.544 12-69 0.0337 126.590 126.50 f 0.03 20 126.54 f 0.05 t 40-0 0.033 f 0.001 Oto-10 0.027 f 0.003 -10 to-20 0.022 f 0.003 -20 to-30 0.017 f 0.005

0.1

130.11 f 0.08

PI,M

Emax (20 "C),

0.1

131.39

0.1 1 x 10-4 0.1 1x 0.1 0.1 0.1

128.56 f 128.47 128.41 f 128.36 127.31 f 127.27 126.60 f

glass-forming compositions), Br-, and I- itself. Samples were prepared by placing a drop of the solution between two General Electric polished fused quartz disks (1 in. diameter, 1/8 in. thickness). These disks were mounted in a Perkin-Elmer liquid absorption cell holder. UV absorption spectra of iodide-containing solutions were obtained by means of a Cary 14R recording spectrophotometer operated at a scan speed of 0.5 nm/s. Temperature control was achieved by placing a 0.25in. thick metal block between the disks and one side of the holder. This block consisted of a brass and copper plate of the same dimensions as the cell holder joined together by tin solder through which copper circulation tubes were passed, making it possible to cool the block by flowing through it dry nitrogen previously passed through a Dewar of liquid nitrogen, or to heat it with water pumped from a Haake FT constant temperature circulator. The other side of the quartz disks was always insulated and mechanically protected by a rubber O-ring pressed between the quartz and the cell holder. The temperature was monitored by means of a Cu-constantan thermocouple, epoxy-glued to the rim of the quartz disks at their junction. The reference solutions were contained in a similar setup without a temperature-control block. Cooling and heating of the samples were always carried out at a slow rate of -1 deg/min to minimize temperature differences across the sample and between the sample and the thermocouple. Using this technique it was found possible to supercool aqueous 0.1 M NaI solution to -31 "C before freezing took place. The thickness of the solution layer in the sample and reference cells was adjusted by tightening the screws of the cell holders. With samples consisting of 0.1 M NaI in water or aqueous solutions of various salts this thickness was adjusted to give an absorbance of 0.3-0.8 at the iodide peak wavelength (near 226 nm). Since Beer's law is still valid a t this concentration range (see below), this corresponds to an optical path length of 2-6 pm and a total sample weight of 4-11 mg. The reference used with samples of NaI solutions in pure water was water, and the reference used with salt-containing solutions was an identical solution without iodide. Acetate and bromide solutions have some background absorption in the spectral region of interest which had to be compensated in any samples containing Mg(OAc)2or NH4Br. This was done by adjusting the thickness of the solution layer in the reference cell until the shape of the

f

0.06 0.04 0.04

0.04 0.04

t 20 to -81

-81 to -132 t 20 to -42 -42 to -94 20 20 20 20 20 20 20

0.035 f 0.005 i 0.037 f 0.005 i

0.001 0.001 0.001 0.001

Ref 1 12 6 This work This work This work This work This work This work This work This work This work This work 1, 13 This work 1, 13 This work 1, 13 This work

iodide spectrum (in particular, the ratio between the intensity of the absorbance at the 226-nm peak and the absorbances at the 193-nm peak and the 208nm minimum between them) became identical with the line shape in pure water solution. In the cases of samples containing high iodide concentrations, the solution layer was squeezed more tightly and peak absorbances as high as 3 were read using neutral density screens. In these cases especially it was found necessary to tighten the screws slowly and uniformly so as to prevent the formation of gaps in the solution film. When such gaps appear in the optical path the iodide spectrum disappears and is replaced by a weak, broad, and ill-defined reflection spectrum from the cavity between the wet quartz surfaces. The spectra of samples containing low concentrations (10.05M) of iodide were measured by using a l-cm Pyrocell quartz cell against identical solutions without iodide. At iodide concentrations between 1 X and 5 X M NaI Pyrocell silica inserts were introduced into the sample (as well m the reference) cell to decrease the optical pathlength to 0.300-0.005 cm. All the determinations of peak absorbances were carried out with the quartz cell. During all measurements both sample and reference compartments of the Cary 14R were purged with dry N2 to prevent condensation and to make it possible to extend the short-wavelength limit of the measurements to 190 nm. Due to the small observed energy shifts, relative to the transition energy, being measured, and to the low signal-to-noise ratio of the spectrophotometer in the spectral region of interest, several spectra were recorded and analyzed to obtain each of the data points presented in the Results section. Each data point represents an average of n measurements, where n = 20 in the solute concentration effect studies, n = 10 in the temperature effect studies on water and n = 5 in the temperature effect studies on Mg(OA& solutions. All the chemicals used were Baker Analyzed reagents. Triply distilled or redistilled deionized water was employed. Glass transition temperatures were measured by means of a Perkin-Elmer DSC-2 differential scanning calorimeter, operated at 10 deg/min, and the densities were obtained by weighing the solutions in 10-ml volumetric flask.

Results 1. Solute Concentration Effects on the Iodide Peak Position. The effect of increasing solute concentration on The Journal of Physical Chemistry, Vol. 81, No. 2, 1977

A. Barkatt and C. A. Angel1

125

3000

I,

Flgure 1. Solute concentration effects on the peak position (a) and the bandwidth (b) of the iodide absorption at 20 OC: (I,.) NaI, A = I-; (0) K I crystals (ref 1 and 5); (II,X) Mg(OAc),, A = OAc-; ( 8 )dry 1:l KOAc-Ca(OAc), glass, A = OAc-; (lo) Li2S04,A = SOP-; (0) MgSO.,, A = SOP-; (A) NH4CI, A = Cr;(0) NH,Br, A = Br-. 0.1 M NaI present in all cases except I. Inset shows Beer’s law test for the iodide absorption, 20 *C.

the position of the lower-energy component of the iodide CTTS doublet is shown in Figure l a and Table I, using units of kilocalories per mole to accord with past practice.’ The results obtained for the dependence of the peak position on iodide concentration between 0.05 and 8.0 M NaI using the quartz disk “sandwich” method show a slight rise of E,,, with [NaI] up to a maximum at 2 M NaI, followed by a decrease below the level of E,, in the case of [NaI] 0. At a concentration of 0.05 M NaI, it was found possible to observe the peak position using both the quartz cell method, with a pathlength of 50 pm and a resulting E,,, = 126.52 f 0.03 kcal mol-l, and the disk “sandwich” method, with pathlengths of 3-4 pm and E,, = 126.54 f 0.05 kcal mol-I. It is concluded that the reduction in pathlength involved in the disk method does Another conclusion, not have a significant effect on E,, which is relevant to the temperature effect studies (see below), is that in 0.1 M NaI E,, shows only a very slight increase, of 0.04 kcal mol-I (equivalent to a change of 1deg in temperature), over its value in the case of [NaI] 0. The effects of other solutes were studied using the disk “sandwich” method. It can be seen that the energy of the transition considerably increases with increasing Mg(OA& concentration, and this increase goes on even at the highest M ~ ( O A Cconcentrations )~ studied here. Higher concentrations, although available up to 4R,15 could not be prepared in suitable thin films for study. A completely anhydrous acetate glass (1:l KOAC-C~(OAC)~, which is 18.14 M in OAc-)14containing NaI was more easily studied and yielded an E,,, of 130.93 kcal mol-1 (Figure la). Although the accuracy in this case is lower than for the solutions due to the high background OAc- absorption the E,, value is certainly lower than the last (12.5 M) solution value. The decrease is believed to be due primarily to the removal of H20 dipoles rather than to the change of cations, implying the existence of a maximum in the E,, vs. concentration plot. An oppositely directed difference is found between the E,, value in the most concentrated

-

-

The Journal of Physical Chemistv, Vol. 81. No. 2, 1977

NaI solutions and that in pure KI crystals’r5 (see Figure la). In order to compare the effects of iodide and acetate with those of other, previously studied, ions a few measurements were carried out on NH4Br,NH4C1,MgS04,and Li2S04,see Table I. These results show that the solute effect obtained using the disk method and 0.1 M NaI is almost identical with that observed previously1J3using 1-cm cells and 1 X M KI for Li2S04,MgS04, and NH4C1. 2. Solute Concentration Effects on Bandwidth and Intensity. The effect of increasing solute concentration on the bandwidth of the iodide absorption is shown in Figure lb. In order to minimize the interference from the 193-nm iodide band and the background absorption of the solutes at low wavelength we measured the separation ‘ I 2 A E between the peak energy and the energy at halfmaximum on the low-energy side of the band. As discussed below, this may be slightly different from the correct half-width obtained by Gaussian resolution of the band but the difference between the two quantities is small and the solute effects measured here can be expected to be valid also for the correct AE. From Figure l b it can be seen that solute effects are generally small, 10.38 kcal mol-’ (6% of the bandwidth), except in the case of NaI. In all cases except >2 M M ~ ( O A Cthe ) ~addition of solutes increases the bandwidth. The order of the solutes with respect to the broadening, relative to M ~ ( O A Cat) ~the composition of comparison, is exactly the reverse of their effectiveness in shifting E,, to higher values, i.e., for broadening the order is NaI >> NH4Br > NH4C1> Mg(0Ac)z > MgS04 > Li2S04. As in the case of the peak position measurements (see above), identical results were obtained for the bandwidths observed in 0.05 M NaI using the cell method and the disk method. 4 M NH4C1, 2 M Mg(OAcI2, 2 M MgS04, and 2 M Li2S04cause a small (1-3%) reduction in the peak absorbance intensity of 0.005 M NaI, measured in the same quartz cell with a pathlength of 0.010 cm. Because of the smallness of the effect it is not given any further consideration. More extensive measurements of the dependence of absorbance on concentration were carried out in the case of NaI itself since it was necessary to find out whether the Beer-Lambert law is valid in NaI solutions up to the concentration range of 0.05-0.1 M and thus is applicable to the determination of layer thicknesses in the disk “sandwich” setup. Using quartz cells, quartz inserts, and neutral density screens it was found that Beer’s law is obeyed within *5% over the entire concentration range studied here, 2 X 10” to 5 X lo-’ M NaI, both with respect to the absorbance at the 226-nm peak, with an e226 of (13.8 f 0.7) X lo3 M-’ cm-’ , and with respect to the absorbance at the 208-nm minimum between the two peaks, with an 6208 of (4.3 0.2) X lo3 M-’ cm-’ (see Figure la, inset). These molar absorptivities, which are in good agreement with previous results,ll6 could therefore be used in the determination of the thickness of layers of 0.05-0.1 M iodide between quartz disks. On the other hand, the assumption that Beer’s law still holds at very high iodide concentrations would lead to the conclusion that in 8 M NaI, where a peak absorbance of 3 was measured, the pathlength was 0.25 pm. In tests it was found impossible to compress films of 0.1 M Na to less than 0.5 pm, and it is therefore probable that a considerable negative deviation from Beer’s law occurs at high iodide concentrations, The deviation may be connected

*

117

CTTS Spectrum of Iodide

Gar..;.. q

21001A

..

5

5.2

,x -125

, -100

-75

-50

-25

TEMPERATURE,

C'

0

I000

25

Figure 2. Temperature effects on the peak position (a) and the bandwidth (b) of 0.1 M NaI in R = 14 and R = 7 aqueous Mg(OAc), solutions. is the temperature of configurational arrest defined by differential scanning calorimetry measurements using scan rates of 10 deg min-'.

with increases in bandwidth observed above 0.5 M NaI. (A similar decrease in intensity takes place, as described above, to a much lesser extent with other solutes.) 3. Temperature Effects. 0.1 M NaI in Aqueous Mg(OAc), Solutions. Measurements were carried out on two glass-forming compositions: R = 14 mol of H20/mol of Mg(OAc), (3.150 M Mg(OAc)2at 20 "C, p = 1.224 g/ml) and R = 7 (4.957 M Mg(OAc)2at 20 "C, p = 1.332 g/ml). The results obtained for the peak position as a function of temperature (see Figure 2a and Table I) show that E,, increases linearly (within experimental error) with decreasing temperature, with a slope similar to dE,,/dT in water above 0 "C, though slightly higher the higher the Mg(OAc), concentration. However, at a certain temperature, -79 f 3 "C in R = 14 Mg(0Ac)z and -41 f 3 "C in R = 7 Mg(OA&, the slope abruptly changes to a value almost an order of magnitude smaller. The temperatures at which the changes occur are very close to the known glass transition ternperature~l~ which have been confirmed within f 1 deg in the present work using DTA and DSC techniques at heating rates of 10 deg/min (see Table I). Due to the limited accuracy of the spectral method, no effects of a changed rate of cooling (in our case, about 1 deg/min), or of differences between heating and cooling were demonstrable. runs around Tg, Bandwidth measurements were carried out on the less concentrated, R = 14 solution for which the background acetate absorption was less troublesome. The results (see Figure 2b) show that the dependence of 'fZAEon temperature is linear within experimental error, with a slope of 0.008 f 0.001 kcal molF1deg-l. The plot agains sharply breaks at T and d(l/,aE)/dT becomes very small (0.001 kcal mol-l deg-') for lower T. 4. Temperature Effects. 0.1 M NaI in Water. The dependence of the peak position on temperature was studied using 2-6-pm films of 0.1 M NaI in liquid water between +40 and -30 "C. The results are presented in

. .

__.*.

_.--

__---__----

58

-20

2050

_/--

2025

40

TEMPERATURE, "C

Figure 3. Temperature effects on the peak position (a) and the bandwidth (b) of 0.1 M NaI in water: (-) ref 6. Inset shows changes in the 225-nm absorbance during freezing and melting: (I) freezing upon cooling at 0.6 deg/min; (11) melting upon heating at 1.0 deglmin.

Figure 3a and Table I. Upon cooling the solution to a temperature between -15 and -31 "C, depending on the type of quartz used, the size of the sample and the presence of scratches or foreign particles, the solution suddenly freezes, as manifested by the disappearance of the iodide spectrum (see Figure 3a, inset), and its replacement by a weak, flat reflection spectrum from the ice layer. Such freezing was observed also in 1-pm films (with 0.3 M NaI to give sufficient absorption). Upon reheating, this spectrum remains unchanged up to a temperature of -4.1 f 0.5 "C, where the iodide spectrum reappears (see Figure 3a, inset). This observation serves as a check of the temperature measurement. Both freezing and melting are completed with 40 s. The appearance of the iodide spectrum is therefore a very sensitive indicator of the phase transition. The results obtained for the dependence of the peak position on temperature in the liquid during cooling and during heating runs are identical within experimental error. They show a linear dependence of E,,, on T between 40 and 0 "C, with a slope equal to that obtained by previous workers.'g6J2 However, a significant curvature is observed below -10 "C and between -20 and -30 "C the slope is found to be less than one-half its high temperature value. The results obtained for the bandwidth as a function of temperature are higher by about 2.6% than those reported by Siano and Metzler6in the 10-40 "C range. This discrepancy is very probably due to the fact that the present results were obtained by direct measurements of the width of the lower-energyhalf of the band, while Siano and Metzler's results were obtained by a more accurate Gaussian analysis which eliminated interference from the 193-nm band and corrected for the very slight skew of the bandshape. However, it can be seen that the two sets of results show an almost identical temperature dependence. Upon extending the temperature range below 10 "C,we observed that the half-bandwidth levels off at about 5 "C The Journal of Physical Chernistw, Vol. 81, No. 2, 1977

118

and retains a constant value of 5.90 f 0.1 kcal mol-’ down to at least -20 “C. Discussion 1. Validity of Thin Film Spectra. The discussion of this work should start with a brief reemphasis of the validity of our thin layer studies. Not only were E, and 1/2aEshown identical in 2-3-pm thin film and normal cell experiments for 0.05 M NaI (Results section) but also (a) complete agreement with previous data on dE,,,/dT is obtained and (b) films as thin as 1 y and containing 0.1-0.3 M NaI show completely normal freezing and remelting behavior. Clearly we are observing bulk solution properties unaffected by surface effects on solution structure. 2. Models for the CTTS Transition, and the Temperature Dependence of the Transition Energy. Toward the interpretation of CTTS spectra the present study introduces an important new element, viz., the separation of vibrational from structural effects on the temperature It will be appropriate therefore to dependence of E,, consider first how our observations on this point can help refine concepts involved in models for the transition. The consistency of the model with observations on solutions will then be reviewed before its application to interpretation of the observations in supercooled water is considered. All models for the transition have as a common concept the excitation of an electron originally located on the ion into a new orbital which extends, to a greater or lesser extent, over a region of solvent molecules surrounding the ion site. In accord with the Frank-Condon principle, the center-of-mass structure of the excited state is identical with that of the normal solvent cage of the iodide ion. The temperature dependence of the transition energy must therefore result from the effect of temperature on this structure and on the consequent change of ground state, and/or excited state, energies. According to the diffuse model: which seems the best available at the moment, it is the change in ground state energy which is of primary importance. The theoretical models differ primarily in the way in which the electron trapping potential established by the solvent cage is assessed, and in the extent to which the electron in the excited state is considered delocalized into the solvent. In all cases the “cage” or‘“cavity” radius is a critical parameter, and changes of the cavity radius with temperature must be invoked if the unusual magnitude of the CTTS energy temperature dependence is to be understood. That the latter concept provides an inadequate basis for discussion is demonstrated by the results shown in Figure 2, which we now discuss. Figure 2 shows what is, with little doubt, true for all solution CTTS spectra, viz., that most (80%) of the temperature dependence of E, vanishes when the solution passes into the glassy state below Tg.16 The association of Tgwith the “freezing” of the primary liquid structure and the loss of all except vibrational and weak, secondary structural, degrees of freedom is well understood by students of viscous liq~ids.l~-’~ Figure 2 therefore shows that the temperature dependence of the CTTS spectra cannot be due to uniform expansion of the solvent cage since uniform expansion can only be achieved by anharmonic vibrations which are virtually unaffected by the occurrence of the glass t r a n ~ i t i o n . l ~ - ~ ~ In the liquid as opposed to the glassy regime most expansion occurs by configurational rearrangements which, according to symmetry- and coordination number-sensitive spectroscopy and computer simulation studies,2°,21usually produce both distortions to lower symmetry of existing The Journal of Physical Chernlstty Voo6 8 7, No. 2, 1977

A. Barkatt and C. A. Angel1

coordination groups and introduction of new groups of lower coordination number. The higher the temperature the greater the range of sizes and distortions of the local coordination groups produced within the solvent and around any solute species. Distortion of a coordination shell should have qualitatively the same effect on the ground state energy as increase of its radius, hence the temperature dependence of the CTTS band maximum is best interpreted in terms of (a) increasing distortions from regular symmetry of existing solvent cages and (b) the generation of new solvent cages containing on the average fewer solvent dipoles to establish the electron trapping potential. These same factors are, of course, also responsible for the increased half-widths associated with decreasing E, values. The oscillator strength du remains approximately constant. Similar considerations must apply to the effect of temperature on the solvated electron spectrum since the ground state for this transition is structurally similar to the excited state of the iodide CTTS transition. (The substitution of the neutral iodine radical in the latter case for the solvent cavity in the former case is of little consequence energetically.) ‘The average radius for the charge distribution for e,;, for instance, is estimated at 2.5-3.0 A in the ground state,2’ while the radius of the excited state for the iodide CTTS spectrum should be 1.25rI-or 2.70 A according to the diffuse mode1.l Unfortunately, while dE,,,/dT for the solvated electron in certain liquids is known (and is large23),no data equivalent to our E,,, vs. T measurement below Tg(temperature independent site distribution) are available.24 3. Solute Effects on E,,, and 1/2AE. Solute effects displayed in Figure 1 establish that the MgtOAc), solutions used for the dE,,,/dT studies are representative of the general case. Figure I shows an order in the solute effectiveness in shifting E,, to higher values, LizS04> MgS04 > Mg(OA& >> NH4Cl > NH4Br > NaI, which is in agreement with previous results (for LizS04,MgS04,and NH4C1) and with the observation17that the solvent effect correlates with the “structure-making” character of the anions. The decrease in E,,, produced by NaI in concentrations above 2 M is unusual but not unprecedented. N-n-Pr4C1 has a similar maximum at 0.5 M, while Nn-Pent4Cl causes E, to decrease below the pure water value even at the lowest concentrations.’ A maximum may also exist in the Mg(OAc)z-HzO system at higher concentrations than those studied here (see below). The solute effects are generally related to the temperature effects. Structure-making solvents, like temperature decreases, decrease the disorder in the solution permitting tighter binding of HzO to I- and thus smaller effective cavity radii. The analogy of solute and temperature effects extends to the relative half-width variations. The composition dependence of the half-width is particularly large in solutions of iodide itself. In this case the change is in the direction expected for a strong structure breaker-indeed, unusually pronounced increases in the diffusion coefficient of water, which depends on disorder, occur in the presence of iodide The bandwidth effect for I- solutions, however, is so large compared with the other solutions examined that direct I- - I- interactions must be suspected. Very pronounced changes, including extrema in E,, and possibly in LLF, must occur in solutions in the ultra-high concentration “hydrate melt” region where there are insufficient water molecules to provide even a single hydration shell since values for E,, in anhydrous acetate glasses (45 800 k 500 cm-’) are below the values observed

119

CTTS Spectrum of Iodide

obtained from studies of Na deposition on vapor-deposited in the most concentrated solutions studied in this work amorphous ice by Bennett, Mile, and Thomas.2a These (see Figure 1). This ultra-high concentration region authors observed a rather flat absorption with a maximum warrants systematic investigation for the insight it should between 40.8 and 48.5 kcal mol-l deg-l. This is supported provide into the relation between electron trapping poby the observation that, in polycrystalline ice in which tentials and dipolar molecule coordination numbers. electrons can be trapped in very low yield, a band is The few available data for solute effects on the abobserved with E, between 44.7 and 46.1 kcal mol-1.29 The sorption spectra of e,; show that 4.6 M MgC12,12 M KF, lower end of this range or the middle of the amorphous and 15 M LiCl shift the e, - absorption in the visible to ice band would, according to extrapolations of the normal higher spectra, and that tkese shifts (as well as shifts water data, be reached at -44.7 "C. caused by a change of solvent1s2,26) have a linear correlation eaq-absorbs in the visible region of the spectrum, and with the shifts to higher energies caused by the same it might be possible to conduct tests of its behavior in the solutes in the case of the iodide CTTS spectrum.26 temperature range 0 to -38 "C using emulsion techniques 4. CTTS Transitions and Structuring in Supercooled to suppress crystallization. The verification of CTTS-like Water. Recently, studies of the physical properties of anomalies in the anomalous temperature region would water have been extended into the metastable supercooled confirm in detail the close relation of the two phenomena region with the finding that both thermodynamic and mass and encourage the extension of CTTS studies to other transport properties exhibit gross anomalies indicative of and structurally interesting systems. ~ some sort of singularity at -45 "C (at 1 atm p r e s s ~ r e ) . ~ ~vitreous Unfortunately, there is little direct structural information Acknowledgment. Support of this research by the in the same temperature interval on which an interpreNational Science Foundation under Grant No. DMR tation of these anomalies can be based. 7302632A01 is gratefully acknowledged. Figure 3 shows that, whereas E,, for the iodide CTTS References and Notes spectrum in Mg(OAc)z solutions varies linearly with M. J. Blandamer and M. F. Fox, Chem. Rev., 70, 59 (1970). temperature over the entire temperature range down to M. F. Fox and E. Hayon, Chem. Phys. Lett., 25, 51 1 (1974). R. Platzman and J. Franck, Z. Phys., 138, 411 (1954). T g(Figure 2) departures from linearity become evident in G. Stein and A. Treinin, Trans Faraday Soc.,55, 1086, 1091 (1959). water below about -10 "C. The decrease in slope becomes M. Smith and M. C. R. Symons,Discufs Faradsy SOC, 24,206 (1957); more pronounced with decreasing temperature down to the Trans. Faraday Soc., 54, 338, 346 (1958). D. B. Siano and D. E. Metzler, J. Chem. Soc., Faraoby Trans. 2, 68, lowest temperature which could be studied before crys2042 (1972). tallization commenced, -31 "C. This unusual behavior, J. Wong and C. A. Angell, J. NonCryst Solids, 7 , 109 (1972). which reflects the temperature dependence of the CTTS (a) C. A. Angell, J. Shuppert, and J. C. Tucker, J. Phys. Chem., 77, 3092 (1973); (b) D. H. Rasmussen and A. P. MacKenzie, J. Chem. transition ground state relative to a slowly varying4excited Phys., 59, 5003 (1973). state, is consistent with the appearance of some constraint (a) J. Hallet, Proc. Phys. SOC.,82, 1046 (1963); (b) K. T. Gillen, D. on the decrease in effective radius of the iodide coordiC. Dougiass, and M. J. R. Hoch, J. Chem. Phys., 57, 5117 (1972). nation shell (i.e., of the "cavity") in the anomalous region. H. Fromherz and W. Menschick, Z Phys. Chem. B, 3, 1 (1929); 7, 439 (1930). A natural explanation would be the restriction on coorD. Meyerstein and A. Treinin, J. Phys. Chem., 66, 446 (1962). dination sites imposed by the rapidly increasing rigidity M. J. Wooten, L. A. Dunn, D. E. Clarke, and H. S. Frank, Nature of the water tetrahedral network. A terminal value of E,, (London),Phys. Sci., 23, 138 (1971). G. Stein and A. Treinin, Trans. Faraday Soc., 56, 1393 (1960). at -127.95 kcal mol-' reached at -45 "C would not be J. A. Duffy and M. D. Ingram, J. Am. Ceram. SOC.,52, 224 (1969). inconsistent with the trend of the data. In this anomalous C. A. Angell and E. J. Sare, J Chem. Phys., 52, 1058 (1970); E. J. region the band half-width stabilizes, also indicative of the Sare, Ph.D. Thesis, Purdue University, 1971. A relatively large d&&d Tabove T,and an arrest of the spectroscopic imposition of some specific distribution of iodide ion sites shift at TBhave been observed in the case of the n a' transition possessing uniform vibrational dynamics. Without more of NOL in the KN03-Ca(N03), glass-forming m i x t ~ r e .These ~ effects detailed quantitative and site-structure-dependent thewere associated with a dependence of L o n the mutual arrangement and interactions between the cations (in particular;Ca*+) and the oretical models for the transition, however, little more can nitrate anion. These considerations do not apply to most intramolecubr be said. transitions. In the case of CTTS the solvent is an integral part of It should be noted, in this connection, that the both the ground and excited states. Accordingly, the d&,ldT observed for aqueous iodide (0.033 kcal mol-' deg- ) is much larger "configuration coordinate" model, adapted from solid state than in the case of NO3' in KNO3-Ca(N0& (0.0138 kcal mol-' deg-l), color center theory for solution studies by Siano and which makes the arrest at T much more pronounced. Furthermore, Metzler; predicts a curvilinear half-width vs. temperature CTTS transitions are allowed and the high molar absorptivity, 13 800 M-' cm" in the case of iodide compared to 7 M-' cm-' for the nitrate relation such as is observed above 10 "C. However, nothing n a' band, makes it possible to use iodie and other species with in this theory limits this prediction to the case of iodide CTTS bands as probes, rather than major components in detecting in pure water (in Mg(OAcIz solutions the observed deT (and other structural changes) in a variety of solvents. d. Kauzmann, Chem. Rev., 43, 219 (1948). pendence is almost prefectly linear). Furthermore an (a) M. Goldstein, J Chem. Phys., 51, 3728 (1969); (b) Trans. N. Y. actual flattening out, as is observed below 0 "C, is not Acad Sci., in press. (Proceedings of the Workshop on the Glass predicted under any circumstances. Transition and Nature of the Glassy State, Dec, 1975.) C. A. Angell and W. Sichina, (Proceedings of the Workshop on the It appears that behavior similar to that which we have Glass Transition and Nature of the Glassy State, Dec, 1975) Trans. observed for the CTTS transition in supercooled water N Y. Acad. Sci. (1976). may also be characteristic of the spectrum of the hydrated W. E. Smith, J. Elrynestad, and G. P. Smith, J. Chern Phys., 52,3890 (1970). electron (the relation of which to the CTTS transition was (a) L. V. Woodcock, Proc. R. SOC. London, Ser. A, 328, 83 (1972); referred to earlier). Although no spectra for eaq- are (b) J. L. Flnney, bid, 319, 495 (1970). available below 4 "C an extrapolation of E,, based on the E. J. Hart and M. Anbar, "The Hydrated Electron", Wiley-Interscience, New York, N.Y., 1970, Chapter 3. linear dE,,/dT observed for T = 4-90 "CZ2predicts that S. Arai and M. C. Sauer, Jr., J. Chem. Phys., 44, 2297 (1966). a value equal to that attributed to ea< in amorphous ice, d€&/dTfor the electron trapped in molecular glasses after lrradiatbn 46.1 kcal mol-', would be reached at only -67 "C. The at low temperatures has been measured during warm-up (H. Hase, latter energy was obtained by the dubious procedure of T. Warashina, M. Noda, A. Namiki, and T. Higashimura, J. Chem. Phys, 57, 1039 (1972); L. Kevan, A&. Radiat Chern, 4,256 (1974)), extrapolating data on methanol-HzO glasses to 100% but this is not an equivalent experiment since thermal annealing with Hz0.27 A more direct observation, which suggests an even changes in the occupied trap distribution occurs. d&,/dTmay be higher temperature for the arrest of E , on cooling, is that almost as large as in the liquid state. The experiment more closely

-

-

The Journal of Physical Chemistry, Vol. 8 1, No. 2, 1977

120

A. S. Kertes and L. Tsimering

rebted to our ghss state observations would be to study the spectrum of the annealed irradiated glass on recooling, when d&ld Twould be much smaller, arid should be comparable to that in Figure 2 below T As far as we know this experiment has not been done for a system wf;ose liquid state spectral behavior is also known. (25) G. Engel and H. G. Hertz, Ber. Bunsenges. Phys. Chem., 72, 808 (1968).

(26) M. Anbar and E. J. Hart, J. Phys. Chem., 69, 1244 (1965). (27) B. G. Ershov and A. K. Pikaev, Zh. f i z Khlm., 42, 2753 (1967); S. Ya. Pshezhetskii, "EPR of Free Radicals in Radiation Chemistry", Wiley, New York, N.Y., 1974, p 86. (28) J. E. Bennett, B. Mile, and A. Thomas, J. Chem SOC.A, 1393 (1967). (29) L. Kevan in "Actions Chimigues et Biologiques des Radiations", M. Haissinsky, Ed., 13 eme serie, Masson, Paris, 1969.

Enthalpies of Mixing and Solution in Trialkylphosphate-Water Systems A. S. Kertes" and L. Tsimering Institute of Chemistry, The Hebrew University, Jerusalem, Israel (Recelved June 9, 1976)

Excess enthalpies of mixing at 25 "C of water with trimethyl-, triethyl- and tri-n-propylphosphate are negative and with tri-n-butylphosphate positive in the concentration range of complete mutual miscibility. Standard heats of solution of water in the esters is decreasingly negative with increasing chain length, becoming slightly positive for water in tributylphosphate. It is postulated that in the ester-rich region the donor strength of the phosphoryl oxygen is main factor for the trend observed. Calculations based on a model involving the lattice theory of mixtures and interchange enthalpies of interacting surfaces support the hypothesis. In dilute aqueous solutions,the trend of increasingly exothermic heats of solution with increasing chain length of the alkyl groups is explained as being due to hydrophobic stabilization of the water structure.

Interpretation of thermodynamic properties of binary mixtures of polar and self-associated liquids is often hampered by the fact that the molecules in the mixtures participate in several simultaneous interactions. The difficulties are illustrated, for example, by the limited applicability of the various statistical thermodynamic treatments based on the model of ideally associated liquids. The critical test for the validity of such models, which assume that the various homo- and hetero-associated species mix ideally, is usually restricted to the mediumdomain composition, in the mole fraction range between 0.2 and 0.8. The predicted excess functions are frequently in a remarkably poor agreement with the experimental data at the two extremes of the mole fraction scale. The models usually fail to bring out the fine features of dilute solutions. Properties of liquid mixtures in which water is one of the components are particularly interesting, and at the same time difficult for interpretation, due to the unique properties of water. Since most of such mixing data involves the whole composition range, but no sufficient data at low concentration ends, we have undertaken to investigate the composition dependence of the thermodynamic properties of such mixtures by measuring integral heats of solution in addition to those of mixing. The obtained data on the present binary systems of water and four trialkylphosphates show that the actual properties of the mixtures are far from simple, and that the oversimplifications inherent in the models advanced might be misleading in these complex systems if not checked against experimental data in the dilute solutions range. In a different context, this report presents a part of a long-term project on a calorimetric investigation of the thermodynamics and thermochemistry of solvent extraction processes. This is the reason for selecting the homologous series of trialkylphosphates, when the highest member of the series, tri-n-butylphosphate, is widely used as a powerful extractant for inorganic compounds from aqueous solutions. As such, its interaction with water has received and continues to receive much attention since the The Journal of Physical Chemistv, Vol. 81. No. 2. 1977

formation of hydrated species affects the extractive capacity of the ester. The polarity of the four esters investigated is essentially the same,' as is their donor strength expressed in donor numbers.' The experimentally determined dipole moments depend on the solvent employed, and range between 2.78 D for trimethyl- and 2.92 D for tributylphosphate in hexane at 20 "C,or between 3.02 D for trimethyl- and 3.10 D for tributylphosphate when carbon tetrachloride is the solvent. The dipole moments of the neat liquids at 25 OC are apparently the same for the four esters at a value of 3.21 f 0.03 D. The differences in the donor numbers are equally insignificant, having the values of 23.0 for trimethyl- and 23.7 for tributylphosphate. As the result of their relatively high dipole moments, trialkylphosphates associate through a system of dipoledipole bonds. Spectral data suggest3 that the extent of dimerization of the esters depends on the solvent employed, and increases with the chain length of the alkyl groups. In hexane at 25 "C,for example, dimerization constants of 0.7, 1.2, and 2.9 dm3 mol-' for triethyl-, trin-propyl-, and tri-n-butylphosphate, respectively, have been r e p ~ r t e d . From ~ recent heats of dilution data4 of these three esters in isooctane at 25 "C, dimerization constants of 0.15,0.13, and 0.21 dm3 mol-l, respectively, have been calculated. The corresponding enthalpies of dimerization, -45.3, -36.7, and -27.5 kJ mol-l, and entropies of dimerization, -167.7, -139.9, and -195.2 J mol-l deg-l, indicate a relatively strong dipole-dipole interaction. The favorable enthalpy changes are more than offset by unfavorable entropy changes during dimerization, thus yielding the small positive free energy changes. The large negative entropy changes reflect a restricted molecular arrangement needed for the dimerization, and the decreasing -AH(dim) values with the alkyl chain length suggests some hindrance due to the bulkiness of the alkyl groups. With the increase of the molecular weight of the esters their mutual solubilities with water decrease. At 25 O C , trimethyl- and triethylphosphate are miscible with water